Argument

Graham Oppy Arguments and Their Defeaters

Intro

Graham Oppy is the hardest atheist to argue with, and the reason is not that his arguments are individually stronger than anyone else's. It is that he has given up the moves that make other atheists easy to beat.

He does not say religious people are stupid. He does not say faith is belief without evidence. He does not say the burden of proof falls entirely on the theist. He does not even claim to have an argument that disproves God, and he says openly that he thinks the arguments from evil fail too. What he claims is narrower and much harder to dislodge: that when you lay theism and naturalism side by side as complete theories of reality, naturalism is the better theory, because it commits you to less and explains just as much.

Everything else he writes is downstream of that. The critiques of the cosmological arguments, the ontological arguments, fine-tuning, and the cumulative case are not independent pillars. They are him clearing away the reasons someone might have for paying the extra theoretical cost.

Which tells you where the fight is. Do not fight him argument by argument. That is ground he has held for thirty years and he knows the literature better than almost anyone alive. Fight him on the one premise the whole structure needs, which is the claim that naturalism explains as much as theism. A theory is only cheaper than its rival if it is buying the same thing.

Concede the parsimony principle immediately and without reluctance. Then ask what naturalism actually explains.

◎ See this page's graph and discover linked topics →

In full

This page catalogs ten of Oppy's strongest arguments, each with the steel-manned form, the defeater, and the follow-up move to expect. Entries 1 to 3 are the core, and if you only prepare three, prepare those. Entries 4 to 8 are his critiques of specific theistic arguments. Entries 9 and 10 are methodological.

Two standing notes on the exchange, both of which matter more with Oppy than with any other opponent.

First, his concessions are real and should be reciprocated. He grants that theists can be fully rational, that atheistic arguments also fail his success test, and that the dispute is about total theory rather than about who is reasoning badly. An apologist who responds to that with the rhetorical toolkit built for popular atheism will look like he has not read him. Match the register.

Second, several of these disputes are genuinely unresolved. The entries say so where it is true. Overclaiming against Oppy is the fastest way to lose an audience that knows his work, and he has the rare quality of being read carefully by people on both sides.

Cheatsheet

The 30-second reply to the whole system: Oppy's case is a parsimony argument, and parsimony only decides between theories that account for the same data. So grant the principle and contest the antecedent. Naturalism does not explain why there is a contingent world rather than nothing, why it is intelligible and mathematically describable, why the constants fall in a life-permitting range, why consciousness exists, or why moral obligations bind. It posits those as brute. Counting entities, theism adds one. Counting unexplained brute facts, naturalism adds many. That is the trade, and the shorter inventory does not settle it.

Fast facts:

• Oppy's central argument is theory-choice, not evidential. He is not arguing God probably does not exist; he is arguing naturalism is the better total theory.
• He explicitly holds that no argument on either side succeeds, including arguments from evil. So he cannot use "the arguments for God fail" as a positive reason for naturalism, and he does not try to.
• His parsimony claim requires that naturalism have equal explanatory power. That premise is asserted far more often than it is defended, and it is where the whole case is vulnerable.
• His brute-terminus symmetry move is his strongest single argument. Prepare it specifically.

Counter-moves:

1. Grant parsimony, contest adequacy. Never dispute the razor. Dispute the tie it presupposes. See Occam's Razor.
2. Count brute facts, not entities. Make him itemise what naturalism leaves unexplained.
3. Press the difference between kinds of terminus. A necessary being explained by its own nature is not the same species of stopping point as an unexplained contingent state.
4. Refuse the argument-by-argument march. Say plainly that you think the individual critiques mostly do not matter, because the case stands or falls on entry 1.

Concessions worth granting:

• Parsimony is a genuine theoretical virtue and theism does posit an additional kind of entity.
• Many theistic arguments are not demonstrations and do not compel assent. Craig-style confidence about the Kalam's premises is stronger than the evidence warrants.
• Oppy's critiques of the ontological arguments are largely successful, and conceding that costs almost nothing.
• The modal intuitions behind the possibility premise are not self-evident, and saying so is honest.

Closing line: "Your whole case is that naturalism buys the same house for less money. I have granted the pricing principle all evening. What I have not granted, and what you have not shown, is that it is the same house. Tell me what naturalism explains about why there is a contingent, intelligible, fine-tuned, conscious, morally-ordered world, rather than what it postulates about it. If the answer is that these are just how the initial state was, then you have not bought the house. You have written 'house' on a piece of paper."

Argument structure

Master objections to the whole page

MO1: "You are treating Oppy as though his only argument is the parsimony one. He has book-length critiques of each theistic argument."

• True, and this page engages them at entries 4 to 8. But the structural claim stands and Oppy would not dispute it: those critiques are defensive. They show that the theistic arguments do not compel. They do not by themselves give anyone a reason to be a naturalist. The positive case is entry 1, by his own arrangement in The Best Argument Against God.
• If every one of his critiques succeeded and entry 1 failed, naturalism would have no positive support. If entry 1 succeeded and every critique failed, naturalism would still be the better theory on his account. That asymmetry is the evidence for where the weight sits.

MO2: "Conceding as much as this page concedes is just surrender."

• Conceding what is true is what makes the remaining disagreement visible. The concessions here are the ontological arguments, the strength of some Kalam sub-arguments, and the reality of parsimony as a virtue. None of them is load-bearing for classical theism, and all of them are things a well-read opponent already knows.
• The alternative, defending every argument at maximum strength, hands him the easiest win available: find the weakest thing you said and spend the exchange there.

MO3: "Oppy says theists are rational. So what is even being disputed?"

• Whether naturalism or theism is true. Rationality and truth are different questions, and one of the useful things about Oppy is that he keeps them apart. Two people can both reason well from different starting points and one of them still be wrong about how things are.

1. Naturalism wins on theoretical virtue

The argument. Theism and naturalism are rival total theories. Compare them as you would compare any two scientific theories, on simplicity, explanatory scope, and fit with the evidence. Naturalism posits one kind of reality, natural causal reality, and accounts for everything we observe. Theism posits all of that plus a supernatural person. Extra commitment, no extra explanatory return. Therefore naturalism is the better theory.

The defeater. Grant the method entirely. Oppy is right that worldviews should be assessed as theories and right that parsimony is a genuine virtue. The argument fails at the hidden premise, which is that naturalism has equal explanatory power.

Three replies, in order of force.

1. Contest the tie before contesting the count. The razor is a tiebreaker, applicable only between theories that account for the same data. So establish first that naturalism explains: that a contingent universe exists at all; that it is intelligible and describable in elegant mathematics; that the constants fall in a life-permitting range; that conscious subjects exist; that moral obligations bind. Naturalism does not explain these. It relocates them into the initial state and calls them brute. That is not explanation, it is placement.

2. Count what parsimony actually counts. David Lewis's distinction matters here. Qualitative parsimony, the number of kinds of thing posited, is a real cost. Quantitative parsimony, the number of instances, is nearly free. Theism posits one further kind, mind, which we already know exists because we are minds. Naturalism posits an initial state that must be extraordinarily specific to yield the observed world, and that specificity is itself a large and unpaid theoretical bill. See Occam's Razor.

3. The simplicity comparison is not obvious in theism's disfavour. Richard Swinburne presses the reverse claim: a hypothesis positing a single person with infinite power, knowledge and freedom is simpler than one positing particular finite quantities, because infinite values are non-arbitrary while any finite value invites the question why that value. On that reading the fine-tuned constants are exactly the arbitrary finite quantities a naturalist must accept as brute. The argument is contested and should be offered as a serious counterweight rather than as a proof.

If pressed: "Explanation has to stop somewhere. You are just demanding explanations I have no obligation to give." That is entry 2, and it is the right place for him to go.

2. Both views end in a brute terminus

The argument. This is Oppy's best single move. Every total theory bottoms out somewhere. Theism bottoms out in God, whose existence and nature are not further explained. Naturalism bottoms out in the initial state of natural reality, likewise not further explained. Both have an unexplained terminus, so theism's terminus buys nothing that naturalism's does not already have. It simply adds a step before stopping.

The defeater. The symmetry is the whole argument, and the symmetry is false. Not all termini are the same kind of thing.

The distinction to press: a brute terminus is one for which there is no explanation. A necessary terminus is one whose explanation lies in its own nature, which could not have been otherwise. These are different, and collapsing them is what makes the symmetry look plausible.

1. The natural world is evidently contingent. It could have been otherwise in countless respects: different constants, different initial conditions, different laws. A contingent thing that could have been otherwise and simply is not explained is a genuine explanatory gap, and the gap is visible precisely because we can specify the alternatives.
2. A necessary being is not in that position. If God exists necessarily, there are no alternatives to specify, and the question "why this rather than that" has no purchase. The terminus is not a gap because there is nothing there to have been otherwise. That is the point of the classical contingency arguments and of Divine Simplicity, on which God's existence is not distinct from His essence.
3. So the two stopping points differ in kind, not just in location. Naturalism stops at something that demands explanation and does not receive it. Theism stops at something whose nature is the explanation. Whether the theistic claim is true is the argument; that it is a different kind of stopping point is not seriously deniable.

If pressed: "'Necessary being' is doing unearned work. Modal claims of that strength are unsupported." Fair, and it is his best follow-up. Two responses. First, he cannot both deny that we have reliable modal intuitions and rely on modal claims elsewhere in his own system, and he does rely on them. See Metaphysical vs Logical Modality. Second, the alternative on offer is a necessary natural terminus, which is a live naturalistic option but a costly one: it requires that the actual constants and initial conditions could not have been otherwise, which is a far stronger modal claim than anything the theist needs, and one no physics supports. See Contingency Argument and Principle of Sufficient Reason.

3. No argument on either side is successful

The argument. A successful argument is one that ought to persuade any reasonable person who does not already accept its conclusion. By that standard no theistic argument succeeds, since reasonable naturalists reject a premise in each without irrationality. Oppy applies the same standard to atheistic arguments and concludes they fail too.

The defeater. Take the thesis seriously and then notice what it does and does not license.

1. The bar is too high to be useful. On that standard almost nothing in philosophy succeeds, and a good deal of science would fail too, since a reasonable person can generally reject some premise of any substantive argument without formal irrationality. A criterion that classifies nearly every argument as unsuccessful is not discriminating between arguments; it is describing a general feature of non-demonstrative reasoning.
2. The thesis is symmetric, so it cannot support naturalism. This is the decisive point and Oppy grants it. If no argument on either side succeeds, that is not a reason to be a naturalist. It leaves the entire positive case resting on entry 1. So the correct response to entry 3 is not to fight it but to accept it and redirect: fine, then everything turns on theoretical virtue, and here is why naturalism does not win there.
3. Replace success with probabilification. Arguments do not have to compel in order to raise probability. A cumulative case works the way evidence works in a courtroom or a Bayesian network: individually inconclusive, jointly substantial. See Bayesian Probability and entry 10.

If pressed: "You have just lowered the standard because you cannot meet it." No. The standard is being identified as inappropriate to the subject matter. Demonstrative certainty is available in mathematics and almost nowhere else, and no one applies the compel-every-reasonable-person test to historical, scientific or moral conclusions they in fact hold.

4. The Kalam's finitist premises fail

The argument. Oppy's Philosophical Perspectives on Infinity is the technical background. He argues the philosophical case against actual infinites does not work: Hilbert's Hotel shows that infinite sets have counterintuitive properties, not that they are impossible; transfinite arithmetic is consistent; and the impossibility of forming an infinite by successive addition assumes what it needs to prove. He separately doubts that Big Bang cosmology establishes an absolute beginning, and holds that even a first cause would not obviously be personal.

The defeater. Concede the parts that should be conceded, then note what survives.

1. The counterintuitiveness arguments are genuinely weaker than their popular presentation. Hilbert's Hotel is a demonstration of strangeness. Treating strangeness as impossibility is not a good inference, and apologists who lean on it hard are overreaching. Say so.
2. The cosmological case does not depend on them. The Contingency Argument requires no finitist premise at all. It asks why a contingent reality exists, and an infinitely old contingent reality is exactly as much in need of explanation as a finite one. Trading the Kalam for the contingency argument costs nothing the theist needs and removes his best-prepared ground.
3. On the personhood of the cause, the inference is not from "first cause" alone but from the character of what is being explained: a cause of the whole spatiotemporal order cannot be spatiotemporal, and the remaining candidates are abstract objects or minds. Abstract objects are causally inert. See Necessary Being is an Intelligent Mind.

If pressed: "So you have abandoned the Kalam." Not abandoned, deprioritised. It is one argument among several, and its finitist premises are its weakest link against this particular opponent. Fighting on your weakest ground against the person who wrote the book on infinity is a choice, not an obligation.

5. Ontological arguments all fail

The argument. Ontological Arguments and Belief in God (1995) surveys the family and concludes none succeeds. Against modal versions, the key point is dialectical: if S5 makes "possibly necessarily exists" entail "necessarily exists," then granting the possibility premise is granting the conclusion, so the premise cannot be a neutral starting point. Parody arguments generate rival maximally-great entities by the same reasoning.

The defeater. Largely concede, and notice how little it costs.

1. The dialectical point is correct. In S5 the possibility premise does the entire work, and it is not more obvious than the conclusion. An honest presentation of the Modal Ontological Argument says this. The codex already does, on the Quinean Rejection of Modal Logic Objection Defeater.
2. Parodies are answerable but the answer is technical, turning on whether the parody predicate is genuinely maximally great or merely great-in-a-respect, and it does not persuade neutrals. Do not stake anything on it.
3. Nothing depends on this. No classical theistic tradition rests on the ontological argument. Aquinas rejected it. Conceding the whole family leaves the contingency, teleological, moral, and historical arguments untouched, and conceding it early buys credibility for the entries that matter.

If pressed: "If your best a priori argument fails, why trust the others?" Because they are different arguments with different structures. A priori demonstration of a concrete existent was always the most ambitious project in natural theology, and its failure is not evidence against a posteriori arguments any more than the failure of a mathematical proof of God would be.

6. The fine-tuning inference is unsound

The argument. Several strands. We cannot assign probabilities over the space of possible physical parameters, because there is no normalizable measure. We do not know that the constants could have been otherwise. A multiverse plus observer selection accounts for the data naturalistically. And the design inference requires a designer whose own specificity is unexplained.

The defeater.

1. Normalizability cuts both ways. If no probability measure can be defined over parameter space, the multiverse explanation is deprived of the same resource, since it needs the parameters to vary across a distribution. An objection that disables both hypotheses does not favour one.
2. The multiverse generator requires tuning. Any mechanism producing a suitable variety of universes must itself be structured in a specific way, which relocates the tuning rather than removing it. See Multiverse and Fine-Tuning Argument.
3. The "who designed the designer" strand is entry 2 again, and gets the entry 2 answer: a necessary being with no parts and no arbitrary parameters is not the kind of thing that has unexplained specificity. See Divine Simplicity.
4. Concede the epistemic modesty point. We genuinely do not know that the constants could have been otherwise. This is why fine-tuning belongs in a cumulative case as one strand among several, not as a standalone demonstration.

If pressed: "You have retreated to a cumulative case because no single argument works." That is entry 10.

7. Naturalism can ground morality

The argument. Moral facts do not require God. Naturalistic moral realism is a live position (Erik Wielenberg is its best-known defender), and evolutionary accounts explain the psychology of moral judgment. Theistic grounding faces the Euthyphro dilemma and does no better.

The defeater.

1. Distinguish grounding from tracking. Evolutionary accounts explain why we make moral judgments and why they take the shapes they do. They do not explain what would make such a judgment true, and if they explain the judgments fully without reference to moral facts, they cast doubt on the reliability of the faculty rather than vindicating it.
2. The queerness problem is an atheist's argument, not ours. J.L. Mackie argued that objective prescriptive facts would be metaphysically odd entities, unlike anything else in a naturalistic ontology, and concluded there are none. That argument does not weaken on Wielenberg's version; brute necessary moral facts standing in no relation to anything else are precisely what Mackie found incredible.
3. Parsimony applies here too, and it applies against Oppy. Positing a realm of brute non-natural necessary moral truths, unexplained and causally inert, is a substantial ontological addition. Theism grounds obligation in the will and nature of a being it has already posited. See Moral Argument.
4. Euthyphro is answered, not dodged, by grounding the good in God's nature rather than in a detached command, which is why the dilemma is false as posed.

If pressed: "Brute moral facts are no worse than a brute God." Entry 2 again, and the same reply: necessity grounded in a nature differs from brute givenness.

8. Divine attributes may not be coherently specifiable

The argument. Describing Gods presses whether omnipotence, omniscience, perfect goodness and the rest can be specified without contradiction or vacuity. If the attributes cannot be stated coherently, theism is not a well-formed theory and cannot be the better one.

The defeater.

1. The classical tradition never claimed univocal specification. Predication of God is analogical, neither univocal nor equivocal, which is Aquinas's position and not a modern retreat. The demand for a fully univocal specification is a demand the tradition explicitly declines, on the grounds that a fully specifiable God would be a finite object among objects.
2. Most paradoxes dissolve under standard qualifications that predate the objection: omnipotence covers what is logically possible rather than the contradictory, and Divine Simplicity blocks the composition puzzles by denying that the attributes are parts.
3. Concede that some formulations do fail. Naive omnipotence and some voluntarist accounts of goodness are indefensible, and abandoning them costs nothing.

If pressed: "Analogical predication is a way of never being pinned down." It is constrained rather than unconstrained: it entails specific denials, such as that God is a body, that He has parts, that He changes, and each of those is falsifiable in principle by a coherent counter-argument. See Anthropomorphism for the same asymmetry applied to biblical language.

9. Theism is not the best explanation of religious experience

The argument. Religious experience is widespread but its content is culturally determined and mutually inconsistent across traditions. Cognitive science offers naturalistic accounts of the disposition to have such experiences. So the phenomenon is better explained naturalistically.

The defeater.

1. A causal account of a faculty does not defeat its deliverances. There is a cognitive-science account of ordinary perception too, and it does not undermine perception. Treating the existence of a mechanism as debunking requires the further claim that the mechanism is insensitive to truth, which has to be argued separately.
2. Cultural variation is expected on theism, since a real object apprehended through diverse conceptual frameworks yields diverse descriptions. Variation in reports is not evidence of no object.
3. The right frame is scope, not adjudication of individual cases. No serious apologist claims a single experience proves God. The claim is that widespread, life-altering, cross-cultural experience of the numinous is one datum among many that theism accommodates naturally and naturalism accommodates by explaining away.

10. Cumulative cases are leaky buckets

The argument. Antony Flew's image, which Oppy presses in a more careful form: piling up individually unsuccessful arguments does not yield a successful one. Ten leaky buckets hold no water.

The defeater.

1. The metaphor is the wrong model of evidence. Buckets are containers whose failures compound; evidence is probabilistic and independent strands compound in the opposite direction. Multiple independent witnesses who are each individually unreliable can jointly establish a fact to a high standard, which is why courts and historians work this way and why nobody thinks that is a fallacy.
2. The relevant structure is a Bayesian network, not a stack of containers. Each argument raises the posterior somewhat; independence is what does the work; and the correct question is whether the strands are genuinely independent. See Bayesian Probability and Cumulative Case for Christian Theism.
3. Flew himself later abandoned the objection, and the reason is instructive: he came to think the design considerations had cumulative force he had underrated.
4. Oppy's own case is cumulative. The argument for naturalism is that it does better across a range of theoretical virtues, no one of which is decisive. He cannot consistently reject cumulative reasoning as such.

If pressed: "Independence is exactly what your arguments lack." A real objection, and it should be met specifically rather than generally. The contingency argument, the fine-tuning of physical constants, the existence of consciousness, moral obligation and the historical evidence for the resurrection do not share a common premise, which is what independence requires.

Tactical opening and closing

Opening. "I want to start by conceding four things, because I think you are right about all four. Parsimony is a genuine theoretical virtue. The ontological arguments do not work. The finitist premises in the Kalam are weaker than most apologists admit. And no argument on either side compels a reasonable person. I think all of that is true, and I do not think any of it helps you, because your positive case does not rest on any of it. It rests on the claim that naturalism explains as much as theism does. That is the one thing I want to talk about."

Closing. "We agreed early that the better theory is the one that explains the most with the least. What I have not been able to get from you all evening is the first half of that. Every time I ask what naturalism explains about why there is a contingent world, why it is intelligible, why the constants are what they are, why anything is conscious, or why anything is obligatory, the answer comes back that these are features of the initial state. That is not a cheaper explanation of those things. It is the absence of an explanation, priced as though it were one."

Live-cite kit

Scholarly, his side:

• Graham Oppy, The Best Argument Against God (Palgrave Macmillan, 2013), the compact statement of the theory-choice case
• Graham Oppy, Arguing about Gods (Cambridge, 2006), the systematic survey
• Graham Oppy, Ontological Arguments and Belief in God (Cambridge, 1995)
• Graham Oppy, Philosophical Perspectives on Infinity (Cambridge, 2006)

Scholarly, replies:

• Alexander Pruss, The Principle of Sufficient Reason: A Reassessment (Cambridge, 2006), on contingency and explanation
• Richard Swinburne, Simplicity as Evidence of Truth (1997) and The Existence of God, 2nd ed. (2004), on why simplicity may favour theism
• Edward Feser, Five Proofs of the Existence of God (Ignatius, 2017), for the classical-theistic terminus argument
• J.L. Mackie, Ethics: Inventing Right and Wrong (1977), for the queerness argument now aimed at naturalistic moral realism

Aphorism: "A theory is only cheaper than its rival if it is buying the same thing."

See also

• Graham Oppy, the person page with his works and commitments
• Occam's Razor, the principle entries 1 and 7 turn on
• Contingency Argument, the argument to trade the Kalam for against this opponent
• Principle of Sufficient Reason, the explanatory principle behind entry 2
• Divine Simplicity, why the theistic terminus has no unexplained specificity
• Naturalism, the position defended
• Cumulative Case for Christian Theism, the method entry 10 defends
• Bayesian Probability, the formal frame for entries 3 and 10
• Modal Ontological Argument and Quinean Rejection of Modal Logic Objection Defeater, entry 5
• Kalam Cosmological Argument and Infinity, entry 4
• Fine-Tuning Argument and Multiverse, entry 6
• Moral Argument, J.L. Mackie and Erik Wielenberg, entry 7
• Metaphysical vs Logical Modality, the modal-epistemology dispute under entry 2
• Atheism, the master hub